The Application Of Splitting Traffic In The Problem Of Grooming Of Dynamic Traffics In WDM Ring Networks

With the development of Wavelength Division Multiplexing (WDM) technology, the capacity of optical link load is increasing. In WDM networks, the traffic grooming technology has been employed to assign traffics to wavelengths in 1998, then it becomes an important and hot problem in the field of optical network. This is also a research topic with high commercial and scientific values. The research of this thesis is focused on ring networks due to the NP-hard property of the grooming problem.The technique of splitting traffic means to divide a traffic into smaller parts or shorter segments so as to fit into a wavelength. By this way, a traffic request is sent to its destination which may pass a few wavelengths, and by sharing the already existing ADMs in the wavelengths, the number of ADM's and wavelengths can be reduced. The capacity of wavelengths can be utilized in a more efficient way at the same time. Applying this technique to traffic grooming will lead to better results.We mainly discussed the application of splitting traffic in the problems of grooming of dynamic traffics in strictly non-blocking and rearrangeably non-blocking way. We proposed the genetic algorithms approach along with local heuristics to deal with this problem. The corresponding results were presented and discussed.The major contributions of this thesis include: (1) To classify the technique of splitting traffic, and give the definitions. What's more, we created a new splitting way; (2) To apply the technique of splitting traffic to the problem of grooming of dynamic traffic; (3) To propose the genetic algorithms approach to realize the grooming of splitting traffic in optical networks.The content of this thesis is as followings:Splitting traffic is a new technique proposed recently. As it can help further reduce the number of ADM's and wavelengths, it will result in optimized grooming results when it is used in traffic grooming problems. We first classified splitting traffic technique and gave the corresponding concepts, then analyzed each kind of splitting traffic technique in detail.We showed a set of MOINLP equations to describe the grooming problems, and based on these equations, we proposed a genetic algorithms approach to solve the strictly non-blocking and rearrangeablynon-blocking grooming problems with dynamic traffics respectively. In the genetic algorithms, splitting traffic technique was applied. It is the first time that this technique is employed in the problem of grooming of dynamic traffic.The corresponding results were given in the thesis, which showed that the splitting traffic technique could lead to optimal results. The discussions on them were also given and the merit of splitting traffic was analyzed.